view release on metacpan or  search on metacpan

lib/FASTX/Reader.pm  view on Meta::CPAN

package FASTX::Reader;
use 5.012;
use warnings;
use Carp qw(confess);
use Data::Dumper;
use PerlIO::encoding;
$Data::Dumper::Sortkeys = 1;
use FASTX::Seq;
use File::Basename;
$FASTX::Reader::VERSION = '1.12.0';
require Exporter;
our @ISA = qw(Exporter);

#ABSTRACT: A simple module to parse FASTA and FASTQ files, supporting compressed files and paired-ends.

use constant GZIP_SIGNATURE => pack('C3', 0x1f, 0x8b, 0x08);


sub new {
    # Instantiate object
    my $class = shift @_;
    my $self = bless {} => $class;
    my $args = {};
    
    # Named parameters: undefined $_[0] will read STDIN!
    if (defined $_[0] and substr($_[0], 0, 1) eq '-') {
      my %data = @_;
        # Try parsing
        for my $i (keys %data) {
            if ($i =~ /^-(file|filename)/i) {
                $args->{filename} = $data{$i};
            } elsif ($i =~ /^-(loadseqs)/i) {
                $args->{loadseqs} = $data{$i};
            } else {
                confess "[FASTX::Reader]: Unknown parameter `$i`\n";
            }
        }
    } else {
      # Legacy instantiation
      ($args) = @_;
    }
      
      
      
    if (defined $args->{loadseqs}) {
      if ($args->{loadseqs} eq 'name' or $args->{loadseqs} eq 'names' ) {
        $args->{loadseqs} = 'name';
      } elsif ($args->{loadseqs} eq 'seq' or $args->{loadseqs} eq 'seqs' or $args->{loadseqs} eq "1") {
        $args->{loadseqs} = 'seq';
      } elsif ($args->{loadseqs} eq 'records') {
        $args->{loadseqs} = 'records';
      } else {
        confess("attribute <loadseqs> should be 'name' or 'seq' to specify the key of the hash.");
      }
    }
    
    $self->{filename} = $args->{filename};
    $self->{loadseqs} = $args->{loadseqs};
    $self->{aux}      = [undef];
    $self->{compressed} = 0;
    $self->{fh}       = undef;


    # Check if a filename was provided and not {{STDIN}}
    # uncoverable branch false

    if ( defined $self->{filename} and -d $self->{filename} ) {
      confess __PACKAGE__, " Directory provide where file was expected\n";
    }

    if (defined $self->{filename} and $self->{filename} ne '{{STDIN}}') {
      open my $initial_fh, '<', $self->{filename} or confess "Unable to read file ", $self->{filename}, "\n";
      read( $initial_fh, my $magic_byte, 4 );
      close $initial_fh;

      # See: __BioX::Seq::Stream__ for GZIP (and other) compressed file reader
      if (substr($magic_byte,0,3) eq GZIP_SIGNATURE) {
         $self->{compressed} = 1;
         our $GZIP_BIN = _which('pigz', 'gzip');
         #close $fh;
         if (! defined $GZIP_BIN) {
           $self->{decompressor} = 'IO::Uncompress::Gunzip';